Improving Uncertainty Estimation With Semi-Supervised Deep Learning for COVID-19 Detection Using Chest X-Ray Images.
Coronavirus
Covid-19
MixMatch
Uncertainty estimation
chest x-ray
computer aided diagnosis
semi-supervised deep learning
Journal
IEEE access : practical innovations, open solutions
ISSN: 2169-3536
Titre abrégé: IEEE Access
Pays: United States
ID NLM: 101639462
Informations de publication
Date de publication:
2021
2021
Historique:
received:
14
05
2021
accepted:
24
05
2021
entrez:
23
11
2021
pubmed:
24
11
2021
medline:
24
11
2021
Statut:
epublish
Résumé
In this work we implement a COVID-19 infection detection system based on chest X-ray images with uncertainty estimation. Uncertainty estimation is vital for safe usage of computer aided diagnosis tools in medical applications. Model estimations with high uncertainty should be carefully analyzed by a trained radiologist. We aim to improve uncertainty estimations using unlabelled data through the MixMatch semi-supervised framework. We test popular uncertainty estimation approaches, comprising Softmax scores, Monte-Carlo dropout and deterministic uncertainty quantification. To compare the reliability of the uncertainty estimates, we propose the usage of the Jensen-Shannon distance between the uncertainty distributions of correct and incorrect estimations. This metric is statistically relevant, unlike most previously used metrics, which often ignore the distribution of the uncertainty estimations. Our test results show a significant improvement in uncertainty estimates when using unlabelled data. The best results are obtained with the use of the Monte Carlo dropout method.
Identifiants
pubmed: 34812397
doi: 10.1109/ACCESS.2021.3085418
pmc: PMC8545186
doi:
Types de publication
Journal Article
Langues
eng
Pagination
85442-85454Informations de copyright
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
Références
IEEE Trans Med Imaging. 2020 Jan;39(1):87-98
pubmed: 31170065
Appl Soft Comput. 2021 Nov;111:107692
pubmed: 34276263
Trends Mol Med. 2020 May;26(5):483-495
pubmed: 32359479
Ann Oper Res. 2021 Mar 21;:1-42
pubmed: 33776178
J Clin Microbiol. 2020 Apr 23;58(5):
pubmed: 32132196
Int J Comput Assist Radiol Surg. 2021 Dec;16(12):2089-2097
pubmed: 34480723
Sci Rep. 2020 Nov 11;10(1):19549
pubmed: 33177550
Sci Rep. 2022 Jan 17;12(1):815
pubmed: 35039620
Phys Eng Sci Med. 2020 Sep;43(3):915-925
pubmed: 32588200
Comput Methods Programs Biomed. 2020 Nov;196:105608
pubmed: 32599338
RNA. 2020 Jul;26(7):771-783
pubmed: 32358057
Radiology. 2020 Apr;295(1):202-207
pubmed: 32017661
Phys Eng Sci Med. 2020 Jun;43(2):635-640
pubmed: 32524445
Wiley Interdiscip Rev Data Min Knowl Discov. 2019 Jul-Aug;9(4):e1312
pubmed: 32089788
World J Surg. 2015 Feb;39(2):363-72
pubmed: 25277980
Quant Imaging Med Surg. 2014 Dec;4(6):449-50
pubmed: 25525575
Cell. 2018 Feb 22;172(5):1122-1131.e9
pubmed: 29474911
Sci Rep. 2017 Dec 19;7(1):17816
pubmed: 29259224
Lancet. 2020 Feb 15;395(10223):507-513
pubmed: 32007143
Radiology. 2020 Apr;295(1):210-217
pubmed: 32027573